2019 Men’s College Basketball Season: Statistics Behind NCAA Tournament Selection

By: Dave Arthur

Questions to Answer

What are the statistics that most seem to determine which teams make the NCAA Division I Men’s Basketball Tournament?
Are teams that succeed in the NCAA Tournament generally better at offense or defense?

“Bubble In” is defined as a 9-11 seed
“Automatic Bid” is defined as a 12-16 seed (i.e. a team that would not have made the tournament without winning their conference tournament)

“Sweet 16” refers to all 16 teams that made the Sweet 16, whether or not they advanced beyond
The other 52 teams that made the tournament are in the “Made Tournament” category

Adjusted offensive efficiency measures how many points a team scores per 100 possessions, and adjusted defensive efficiency measures how many points a team gives up per 100 possessions
NCAA Tournament teams have both good offense and good defense (few teams outside the bottom right quadrant of the graph)
The teams that succeeded in the NCAA Tournament by making the Sweet 16 or better all had very good offense and very good defense
For the best teams, there is not a strong indication that offense or defense is more important than the other
Virginia, which won the NCAA Tournament, had slightly better offense relative to other teams than defense

The most correlated variable with making the NCAA Tournament is WAB (wins above bubble), with a correlation of 0.663. This statistic measures how many more wins a team has against their schedule than the average “bubble” (a team on the cut line of the tournament) team would.
Other strongly positively correlated statistics include W (wins), BARTHAG (an estimate of how likely a team would be to win against the average team), ADJOE (adjusted offensive efficiency), and G (# of games played). As teams rate higher in these categories, they are more likely to make the tournament.
Strongly negatively correlated statistics are defensive statistics like ADJDE (adjusted defensive efficiency), EFG_D (effective field goal percentage on defense), and X2P_D (defensive 2 point rate). As teams have lower values in these statistics (i.e. as they better minimize their opponents’ abilities to score), they are more likely to make the tournament.
ADJDE has a correlation of -0.520, and ADJOE has a correlation of 0.556. This furthers the hypothesis that offensive and defense are roughly equally important when determining tournament-quality teams.

Wins above bubble is a very heavy factor in terms of which teams make the tournament and where they are seeded once they do so
Clear distinction between WAB of top seeds, mid-level seeds, and bubble teams
WAB also seems to predict well which teams succeed in the tournament
However, upsets do happen and it’s thus not a perfect indication (as evidenced by the small green hump on the left side of the right graph, which is 12 seed Oregon)
We should also note that 2019 was an unusually chalky year for the tournament, with Oregon being the only team seeded worse than 5 to advance to the Sweet 16
We should keep in mind that correlation does not imply causation, and the WAB statistic is not the sole arbiter of which teams do well
Better-seeded teams generally perform better in the tournament both because the teams have to be better to earn those seeds and because the bracket is set up in a way such that teams with better seeds end up playing teams with worse seeds. The similar patterns in WAB density for the seeding chart and the performance chart reflect that teams with better seeds tend to win more games.

Keep in mind that only one year was analyzed, and specific data regarding correlated statistics may vary from year to year
In 2019, offense and defense seemed to be equally important both for tournament selection and tournament performance, with perhaps a slight edge of importance given to offense
Wins Above the Bubble was the most correlated statistic with tournament selection and was also a good predictor of which teams succeeded in the tournament by making at least the Sweet 16
Other statistics like the number of wins, BARTHAG, adjusted offensive and defensive efficiencies, and the number of games played (which makes sense given postseason games also counted towards that statistic) were other statistics strongly correlated with tournament selection
The WAB correlation is interesting given that, unlike other analytical data, the NCAA Tournament has never used it in the selection process. However, they are adding WAB to the selection process this coming year, so it will be interesting to see if there is a sort of positive feedback loop resulting in WAB becoming even more correlated with tournament selection.